Proteomics of Mouse Models for Human Disease. Proteomic studies were performed using mouse models in order to gain a further understanding of the pathophysiology for two human diseases: a) Niemann-Pick Disease-Type C1 (NPC1), and b) tumor aggressiveness of pheochromocytomas/paragangliomas. In each model system, proteins were separated by 2-dimentional gel electrophoresis (2D-GE) and quantified by the intensity of spot staining. Differentially expressed protein spots (> 1.5 fold change, p<0.05) were then excised and identified by mass spectrometric techniques. A) NPC1. Mouse cerebellum tissues from Npc1 knock-out and wild type animals at 1, 3, and 5 weeks of age were compared. Protein identities were obtained for 77 differentially expressed proteins. These proteins include those involved in glucose metabolism, detoxification/oxidative stress, and Alzheimer disease-related proteins. Members of the fatty acid binding protein family, FABP3, FABP5, and FABP7 were also found to have altered expression. Translating these findings to CSF from NPC1 patients, glutathione-S-transferase a, superoxide dismutase, and FABP3, were found to have altered expression relative to control CSF. B) Protein expression in aggressive and non-aggressive mouse tumor cell lines developed from mouse PHEO cells. Fifty-three differentially expressed proteins were observed and identified by mass spectrometry. Ingenuity pathway analysis indicated that the OXPHOS pathway was the pathway with the most significant changes, with four spots identified as subunits of OXPHOS complexes, including ubiquinol-cytochrome-c reductase protein 1, and ATP synthase &#946;, &#947;, and &#948; subunits. Expression of the glycolysis related protein, lactate dehydrogenase B, decreased in the aggressive tumor cell line, while expression of lactate dehydrogenase A and SOD2 increased. The later two proteins have been identified as promising therapeutic targets in other cancers. Protein Profiling and Quantification in Cerebral Spinal Fluid (CSF) for Biomarkers of NPC. Initial experiments on protein profiling demonstrated detection of significant differences between proteins in CSF from NPC patients and adult controls using a novel analytical approach that we developed. Protein profiles were compared using MALDI TOF (Matrix Assisted Laser Desorption Ionization-Time Of Flight) mass spectrometry and the combined use of ANalysis Of Variance coupled with Principal Component Analysis (ANOVA-PCA). In addition to the application of MALDI-TOF profiling, we have begun developing a method to compare relative levels of specific proteins in CSF from patients, using an 8-plex iTRAQ labeling process to mass tag peptides generated from proteins present in each sample. The method being developed will perform relative protein quantification while maintaining individual patient information and providing the ability to compare results across multiple iTRAQ experiments. Initially, studies in the Npc1 knock-out mouse model are being used to determine the analytical variation of this method. Several proteins identified in our pilot study were also identified in our previous study of differentially expressed proteins in the cerebella of via 2D-GE. One example of a differentially expressed protein detected by both methods was ATP-synthase alpha which showed a 2.6-fold decrease (p=0.002) via 2D-GE while initial iTRAQ experiments resulted in 1.7-fold decrease (p=0.07). An additional protein that was found only in the iTRAQ experiment was Purkinje Cell Protein 4, which was down-regulated in the mutant mouse 1.9-fold (p=0.02). Purkinje cells are known to be the first neuronal cell type to succumb in NPC1 disease. After completion of the variability study, this method will be used to profile protein changes in CSF from patients with the fatal, neurodegenerative, Niemann-Pick Disease, type C1. Characterization and Quantification of Serum Cardiolipins. We have developed a methodology to extract and quantify cardiolipins in human serum. This effort is associated with a clinical study in the Institute to evaluate the effects of antibiotic treatment for pregnant women colonized with Group B streptococcal (GBS) organism on serum cardiolipins. The hypothesis of the study is that the typical peri-natal penicillin treatment gives rise to a large increase of circulating cardiolipins in the infant, which then leads to respiratory distress. The analytical approach involves addition an internal standard to a serum sample, followed by liquid-liquid and solid phase extractions. The cardiolipins are then detected and quantified by LC-MS analysis. Using this method, we have analyzed cardiolipins in normal adult serum, and determined concentrations of (18:2)-4 cardiolipin in a range between 3 -10 nM, approximately 1000-fold lower than reported by earlier, less accurate measurements.. In addition, cardiolipins with other fatty acid moieties have been detected. This method is currently being applied to quantify cardiolipins in maternal serum, cord blood, and infant serum from the clinical study. New Approach to Mass Spectrometric-Based Protein Identification. Current approaches to protein identification rely on database matching of fragmentation spectra, while ignoring (in a scoring sense) the mass accuracy of the precursor ions. We have developed a method that uses targeted peptide mass fingerprinting to confirm MS/MS database search identifications. The distribution of mass errors of matches in the first order spectrum is used to develop a probability model that is independent from MS/MS database searches, employing Bayesian Inference and the mass errors in the first-order spectra to assign peptide scores. Products of the peptide probabilities yield the final protein scores. These true protein scores are then assigned a probability of being correct by applying Extreme Value Theory to a set of randomly selected decoy proteins that are approximately the same mass as the true proteins, but with scrambled sequences. The method was tested on a mixture of protein standards and results demonstrated both very high confidence levels for proteins identified by MS/MS fragmentation, and also validated several low scoring one hit wonders found by MS/MS fragmentation. Mass Spectrometric-Based Profiling of Urinary Steroids. Current approaches to the analysis of urinary steroids typically employ either immunoassay or mass spectrometry based technologies. Immunoassay-based methods often lack specificity due to cross-reactivity with other steroids, and targeted LC-MS/MS is limited to the analysis of pre-determined analytes. We have developed a new LC-MS/MS approach to urinary steroid profiling that enables us to detect the steroids that have truly changed in a patient cohort without knowing their identity beforehand (i.e., untargeted metabolomics of steroids). In addition, we have utilized a product ion spectrum database of known steroids to improve our capability to identify novel steroids. These methods were tested in a pilot project to investigate urinary steroid for patients diagnosed with polycystic ovarian syndrome (PCOS). Initially these studies detected elevated levels of an unknown compound consistent with an androgenic steroid in PCOS patients. We were then able to identify the unknown as a mixture of androsterone-sulfate and etiocholanolone-sulfate. Two dimensional liquid chromatography, using C18 in the first dimension and Porous Graphitic Carbon in the second, was required to separate the diastereomeric pair.